VirtConsole 3.3.7 makes every backup something you can trust and prove. Backups of running instances are taken with the guest’s file systems frozen for an instant, every local, S3 and rclone backup carries a checksum, and you can restore or drill any backup destination, not just Proxmox Backup Server.
Consistent backups of running instances
Before the point in time of a backup is set, the agent asks the guest agent to freeze the guest’s file systems, then thaws them right after. Windows guests use VSS through the same agent. The copy runs after the thaw, so the freeze lasts for an instant.
- Each backup records its consistency: file-system consistent, or crash consistent when no agent answers or the freeze takes too long.
- The backup completes either way, and the consistency shows on the backup row and in the admin API as
consistency. - It applies to local, S3, rclone, Ceph and Proxmox Backup Server backups of running instances.
SHA-256 checksums on every backup
Local, S3 and rclone backups, full and incremental, on qcow2, raw block and Ceph disks, are hashed as they are written.
- The checksum is recorded with the backup and shows as
checksumon the admin API’s instance backup rows. - Before a restore uses each backup in a chain, the node checks it against its recorded checksum.
- Backup chains also re-anchor cleanly after an interruption or a change of backup destination.
Restore to a new instance, for every destination
Admins can restore local, S3 and rclone backups into a new instance, including a whole incremental chain. A local backup restores on the node that holds its files. S3 and rclone backups restore on any KVM node, so you can bring an instance back on a different node or in a different location. Proxmox Backup Server restores work as before.
Read the backups guide.
Recovery drills for local, S3 and rclone backups
A recovery drill covers local, S3 and rclone backups as well as Proxmox Backup Server.
- The chain is restored into a disposable guest with no network, boot-checked through the guest agent, timed and destroyed.
- A passed drill marks each drilled backup verified, based on the checksums the node checked during that restore.
- Ceph snapshot backups are recorded as a skipped drill.
- A drill row names its destination in the admin API as
backup_destination_type.
Read the recovery drills guide.
A health dashboard that clears itself
The Cron, Queue and Mail tiles on the admin dashboard return to healthy once the same scheduled command, job class or mailer has run healthily ten times in a row after a failure.
- Queue and Mail keep the last three failure messages for you to read.
- Health tracking adds no database write per job.
Read the dashboard guide.
Thanks for building on VirtConsole. Questions along the way, reach us any time.